Data for: Evaluation of pollination traits important for hybrid wheat development in Great Plains germplasm
Data files
Apr 05, 2023 version files 291.80 KB
-
Raw_Data.xlsx
-
README.md
-
Trait_Description.xlsx
Abstract
Hybrid wheat (Triticum aestivum L.) offers potential yield advantages over conventional inbred cultivars. For hybrid wheat to be a commercial success, the cost to produce the hybrid seed needs to be minimized. Although wheat is naturally self-pollinated, hybrid wheat seed production can be improved by increasing the amount and availability of pollen for cross-pollination. This research examined 19 pollination traits using the Hard Winter Wheat Association Mapping Panel for three years. Anther extrusion, pollen 50 date (date at which a genotype has 50% of spikes pollinating), plant height, and pollination duration (last spike pollen 50 date minus first spike pollen 50 date) were identified as the most important traits for hybrid seed production. Anther extrusion, plant height, and pollen 50 date varied widely among genotypes, while pollination duration had significant genotypic differences in one year of testing. These traits also had significant genotype x year interactions, but better and poorer performers were consistent among years. Anther extrusion was weakly, negatively correlated with plant height, and high anther extrusion semi-dwarf genotypes were identified. Pollination duration was reduced in a high temperature (>30ºC) environment, and genotypic differences in pollination duration were identified only in a milder temperature (24ºC) environment. Hierarchical clustering suggested that excellent pollinator genotypes with high anther extrusion and longer pollination duration tended to pollinate early and were of short to moderate stature. Pollination traits were higher when temperatures were mild, which benefited early genotypes because they pollinated before higher temperatures limited their pollination duration.
Methods
The Hard Winter Wheat Association Mapping Panel (HWWAMP) consists of 299 historic, experimental (pre-2010), and released genotypes from both public and private institutions is a valuable genetic resource for studying marker trait associations in Great Plains genetic backgrounds (Guttieri et al., 2017). The HWWAMP was grown at the University of Nebraska Lincoln Agronomy Farm (40º51’23.3”N, 96º36’23.8”W) in 2016, 2018, and 2019. Genotypes were planted in 1m long plots with 30 cm row spacing: as single rows in 2016 (harvest year with the planting being in the previous fall) and in two-row plots in 2018 and 2019 using a Wintersteiger Rowseed Planter (Salt Lake City, U.S.A.). Planting population was 2.5 million seeds per hectare, based on the recommended seeding rates for the area. Each trial was intensively managed so as to remove the effect of disease and was treated with [2-[[[1-(4-chloropheny)-1H-pyrazol-3-yl]oxy]methyl]phenyl]methoxy-, methyl ester), metconazole, 5-[(4-chlorophenyl)methyl]-2,2-dimethyl-1-(1H-1,2,4-triazol-1-ylmethyl) cyclopentanol Metconazole plus pyraclostrobin (TwinLine®, BASF Crop Protection, Research Triangle Park, NC) at Zadok’s 45 and 5-[(4-chlorophenyl)methyl]-2,2-dimethyl-1-(1H-1,2,4-triazol-1-ylmethyl) cyclopentanol Metconazole (Caramba®, BASF Crop Protection, Research Triangle Park, NC) at Zadok’s 50 to control fungal diseases. Fungal diseases were controlled as commercial hybrid seed production would be optimized to have the highest grain yields possible. A modified augmented experimental design was used with an incomplete block size of 24 genotypes and three repeated check genotypes (‘Camelot’, Baenziger et al., 2009; ‘TAM 107’, Porter et al., 1987; and ‘NE06545’, Baenziger et al., 2014), per incomplete block. Two full replicates were planted in 2016 and 2018 and one full replicate was planted in 2019.
Description of Traits Measured
In total, 19 traits (anther extrusion, first spike heading date, heading date, last spike heading date, first spike pollen 50 date, pollen 50 date (syn. anthesis date, when 50% of the spike in a plot were pollinating), growing degree days accumulated, day length at pollen 50 date, photothermal accumulated units, last spike pollen 50 date, spike emergence duration, first spike heading date to heading date, heading date to last spike heading date, pollination duration, first spike pollen 50 date to pollen 50 date, pollen 50 date to last spike pollen 50 date, heading date to pollen 50 date, and total flowering period) were measured or calculated (Supplemental Table 1, Supplemental Figure 1). These traits are a comprehensive set with the idea that we wanted to measure many traits and then, based upon our results, determine which traits had no genetic variation or were correlated such that collecting data on one trait effectively meant we did not need to collect data on the correlated trait. Our goal was to develop reduced list of important traits for hybrid wheat male parent selection. We did not want to assume a trait was important or unimportant until we had results to support this conclusion. Anther extrusion (visually scored where 1 was no anthers extruded and 9 was where all the anthers are extruded was measured shortly after the pollen 50 date when the majority of plants would have extruded anthers), plant height, pollen 50 date, growing degree days accumulated (GDD), day length, and photothermal accumulated units were measured in all three years. The remaining traits were measured or calculated in 2018 and 2019 (Supplemental Table 1, Supplemental Figure 1). First spike heading (syn. head emergence) date and last spike heading date were recorded when the spike emerged from the boot, and the spikes were marked with scientific tape. Those marked spikes were then used to estimate when primary and later tillers pollinated to determine first spike pollen 50 date and last spike pollen 50 date. However, we observed that the specific spikes to emerge first/last were not always the first/last to pollinate. Growing degree days accumulated were calculated with temperatures retrieved from nearby weather stations on a given genotype’s pollen 50 date, with a start date of January 1, and Tbase (growing degree day base temperature) of 0ºC (McMaster and Wilhelm, 1997).
Day lengths were retrieved from https://timeanddate.com (verified Dec. 8, 2021). Photothermal accumulation was calculated by multiplying the GDD accumulated by day length when the genotype reached pollen 50 (Li et al., 2018).
Statistical Analysis
ASREML 4 for R was used to calculate mean-adjusted Best Linear Unbiased Predictions (BLUPs) for each trait (Butler et al., 2018). All traits were run through a model selection pipeline in R to evaluate and correct for spatial variability (Supplemental Materials 1) using standard mixed linear models and mixed linear models with various variance-covariance structures (Belamkar et al., 2018). The best model was then selected based on Akaike information criterion (AIC), Bayesian information criterion (BIC), and visual diagnostics of residual plots for each trait/year combination, and a best overall model was also selected for trait x year (e.g. environment) testing (Belamkar et al., 2018). Trait x year testing was conducted using a cross-classified model as fitting an unstructured or factor analytic model yielded similar results (Isik et al., 2017; Supplemental Materials 2). The selected models were then used to generate mean adjusted BLUPs for each individual trait x year combination and an overall mean adjusted BLUP for each trait. Likelihood ratio tests were used to determine both genotype x year (within cross-classified model) and genotype significance (within the best overall model) by removing either the genotype x year term or the genotype term and testing if the models differed significantly using the lr function from asremlPlus R package (Brien, 2019). Repeatability (individual year) or broad sense heritability (multiple years) was estimated with the two equations given in the associated article.
Multivariate Analysis
Multivariate analyses were conducted using the mean-adjusted BLUPs in R 3.5.2 calculated above (Belamkar et al., 2015; RStudio Team, 2019). A description of R functions and packages used for multivariate analyses can be found in Supplemental Materials 3. Spearman rank correlations using the original BLUPs were calculated to examine trait x year interactions to determine if phenotypic ranks changed. The phenotypic diversity of the HWWAMP along with the presence of sub-groups based on phenotype were investigated through hierarchical cluster analysis and principal component analysis (PCA). Both analyses were conducted using data standardized to mean = 0, standard deviation = 1. The Euclidean distance matrix was generated using the standardized dataset with the intercluster distance estimated using Ward’s linkage method with the grouping cutoff set at 100 (Ward, 1963).
Genetic correlations were calculated using BLUPs for each genotype and trait on standardized data (mean=0, standard deviation =1) after accounting for effects of year and experimental design using SAS v9.4 (SAS Institute, Cary, NC, Supplemental Materials 4). The phenotypic diversity of the HWWAMP along with the presence of sub-groups based on phenotype were investigated through hierarchical cluster analysis (Ward’s method) and principal component analysis (PCA). Both analyses were conducted using BLUPs from standardized data (mean = 0, standard deviation = 1) for the 11 traits with statistically significant genetic variance. A two-dimensional PCA plot comparing Principal Component (PC) 1 vs PC2 was used to investigate phenotypic diversity and the traits that contribute to the phenotypic diversity while also validating the findings from the hierarchical clustering
Usage notes
ASREML, SAS, JMP